This type of game is based off a dataset that came out a few years ago, that was originally built to train AI. It’s built by analysing all text fed to it (pretty much all of Google book and a very large part of the internet) and whether a word is close or not depends on how frequently the words appear near each other.
That’s what I thought for yesterday’s puzzle and tried “final” and “version” but apparently there was no connection between them. I guess if you keep playing you’ll start learning the dataset too.
This type of game is based off a dataset that came out a few years ago, that was originally built to train AI. It’s built by analysing all text fed to it (pretty much all of Google book and a very large part of the internet) and whether a word is close or not depends on how frequently the words appear near each other.
That’s what I thought for yesterday’s puzzle and tried “final” and “version” but apparently there was no connection between them. I guess if you keep playing you’ll start learning the dataset too.
One thing I misunderstood is that it’s not about how close they are semantically, but in meaning. It’s subtle.
I just played for the first time and got:
Game #144
Shortest path: 3 (Avg. 3)
Total words: 3 (Avg. 3)
🟦🟪🟥 | 🔥 1
I must have just got lucky with the word I chose: >!food!<
Interestingly, I chose the same word that you did. But the direct connection to “living” seems to be a stretch.